If a store hikes the 'before' price first, is your personalized discount real savings or just a disguised markup?
Can list-price inflation undermine the meaning of personalized discounts?
This explores whether a seller can make a 'personalized discount' meaningless, or even turn it into a disguised markup, by first raising the list price the discount is measured from.
This explores whether a seller can empty a 'personalized discount' of meaning, or turn it into a hidden markup, by first raising the list price the discount is measured from. The corpus says yes, in principle. The sharpest version of the argument comes from critics of Google's personalized pricing tool for retailers. Google promised the tool wouldn't raise prices. Can Google's pricing tool really prevent price increases? argues that promise can't be tested: if merchants can set any reference price they like before discounts apply, then 'no price increases' only constrains the number nobody actually pays. Is Google's personalized pricing actually surveillance pricing? goes further and calls it surveillance pricing, meaning premiums dressed up as discounts. Both point to the same precedents: hospital chargemaster bills, pharmacy markups, airline fares and department-store 'sales'.
One caveat up front. Both of these are arguments by analogy to healthcare and retail billing. Neither shows consumers actually overpaying under Google's tool. The corpus has the mechanism but not the measurement.
The less obvious part comes from the AI-shopping research, which suggests the problem gets worse when software does the buying. A discount only means something if the buyer can tell what a fair price is. Seller-side agents are getting very good at working that out about the buyer: Can a seller infer what a buyer really wants to pay? found a seller agent could recover a shopper's target willingness to pay almost perfectly from conversation alone. Pair that with an inflated list price and the seller can set the 'discount' so the final price lands just under your personal ceiling. It feels like a deal, but it's priced at the most you'd accept.
The buyer side looks weak against this. Do shopping agents respond more strongly to price and rating cues? found LLM shopping agents react to price and nudge cues far more sharply and mechanically than humans do. That is exactly how a buyer behaves when a large 'X% off' anchor drives the decision. Do LLM negotiation agents reliably follow budget and pricing rules? found buyer agents accepting inflated prices and revealing their budgets without being asked. If the agent working for you takes the crossed-out price at face value, the inflation trick works on it more reliably than it would on you.
Who runs the system matters too. Do advertising-funded shopping agents favor paid product placements? warns that platforms paid by advertisers have built-in reasons to steer shoppers. Do platforms and sellers exploit AI shoppers together more effectively? shows that when a platform and its sellers each exploit the same AI biases, without coordinating, consumer harm more than doubles. Inflated reference prices fit that pattern: a discount badge looks like a description detail, but it works as an anchor. The takeaway: a personalized discount tells you nothing unless the base price is independently checkable, and right now neither regulation nor AI shopping agents reliably check it.
Sources 7 notes
Stoller argues the promise is hollow because merchants can set arbitrarily high reference prices before discounts apply, analogous to inflated hospital bills and pharmacy markups. The claim rests on discount-washing patterns in healthcare rather than direct evidence of consumer overpayment.
Doctorow argues Google's personalized pricing facility sold to major retailers functions as surveillance pricing, extracting premiums under the guise of discounts—a tactic already mastered by airlines, medical billing, and department stores.
A seller-side inference agent recovered a consumer's target willingness to pay with a slope of 1.00 from dialogue transcript alone, when the buyer inhabited a verbal persona. A numeric budget kept confidential produced only a 0.21 slope, showing behavioral coherence with a persona is far more legible than explicit numbers.
A benchmark testing 17 LLM models against humans on binary product choices found agents exhibited much larger choice shifts in response to price, rating, and nudge interventions—collapsing into rule-like patterns rather than the modest probabilistic shifts humans showed.
Across GPT, Qwen, and DeepSeek models, buyer agents disclosed budgets unnecessarily, accepted inflated prices, and ignored cost limits, while sellers accepted below-wholesale offers. All four anomalies traced to instruction-following failures rather than strategic behavior.
Show all 7 sources
Bob Hedges, former Visa Chief Data Officer at MIT, warns that advertising business models create structural incentives for shopping agents to favor paid placements over neutral recommendations, independent of technical design quality.
A multi-agent simulation shows that when platforms manipulate ranking and sellers manipulate descriptions independently, their joint exploitation produces consumer harm exceeding double what each strategy causes alone. This occurs without coordination because both target identical AI cognitive biases.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- When Agents Shop for You: Role Coherence in AI-Mediated Markets
- Algorithmic Collusion by Large Language Models
- Are Behavioral Pricing Tactics Dying? A Study of Price Endings & Partitioned Pricing Effects on Machine Buyers
- Vertical tacit collusion in AI-mediated markets
- A Framework for Studying AI Agent Behavior: Evidence from Consumer Choice Experiments
- What Is Your AI Agent Buying? Evaluation, Biases, Model Dependence, & Emerging Implications for Agentic E-Commerce
- The Automated but Risky Game: Modeling and Benchmarking Agent-to-Agent Negotiations and Transactions in Consumer Markets
- LLM-Agent Interactions on Markets with Information Asymmetries